“The Bursts of Anti-Robredo Posts Partially Driven by Groups with Short-lived Political Activity” is a study presented by the author from ACCeSs@AIM in the eleventh #FactsFirstPH research briefing held on May 6, 2022. The full copy of the research is reposted with permission.
Almost all internet users in the Philippines use Facebook. Because of this, Facebook has become an effective way of propagating information.
However, the ease of sharing information on Facebook is a double-edged sword. On the one hand, users can connect with their loved ones, share educational content, or ask for donations to help victims of a natural disaster. On the other hand, it can be used to spread disinformation or “fake news” that can mislead users into making wrong conclusions and decisions.
One of the worst cases where Facebook was maliciously used was in the Cambridge Analytica scandal that turned the Philippines into a “petri dish” for testing algorithms to manipulate voter opinion.
As the presidential election draws near, we aimed to investigate trends in political posts in Philippine Facebook groups. We analyzed around 24 million posts from 8,900 groups that Rappler collected from January 1, 2019, to September 30, 2020. We looked at trends of political activity and sentiment or attitudes over time, focusing on posts about Robredo, Marcos, and Duterte.
Trends indicate synchronous bursts of political activity
Shown above are positive (green) and negative (red) posts containing keywords about Duterte, Robredo, and Marcos. Marcos- and Duterte-related posts were generally more positive than negative across time. Marcos had far more positive sentiment than negative sentiment. Robredo-related posts, meanwhile, were more negative than positive, especially in 2019. However, in 2020, there was an increase in the number of positive posts over negative ones.
Particular to Robredo-related posts were “bursts” of negative posts as shown in the two red peaks in November 2019 and January 2020. These indicate synchronous activity when many negative posts were being posted over a one- or two-month period. These bursts of negative posts were not present in Duterte- and Marcos-related posts. To analyze these groups further, we looked into the trends of political activity.
Some Facebook groups became more political through time
We categorized the Facebook groups based on the trends of the proportion of political posts over time. We defined a high political proportion as at least 40% of all posts in any given month, while a moderate proportion as 10-40%, and a low proportion as less than 10%. We focused on groups that had high political proportions or activity, constituting 3,175 groups out of 8,900. Among these groups, 60 had an upward trend, 1,242 groups had a stable trend, and 1,873 had a downward trend. We focused on the last two types of groups for this analysis, which we will call, “stable” and “unstable” groups.
The charts below show the changes in the overall proportion of political posts over time for each type of group. These are plotted against each group’s relative month, which is the month since the first data was collected from that group.
These trends indicate the growth and stability of political activity in a group. Stable proportions indicate a stable political group, maintaining a high political activity over time. A downtrend indicates less stability where high political activity was not sustained over time.
Pro-Marcos and pro-Duterte shifted from stable groups to unstable groups
We show above the proportion of positive posts that came from stable groups (green) and unstable groups (blue) per week. Robredo-related positive posts consistently came mostly from stable groups. Meanwhile, Duterte- and Marcos-related posts coming from stable groups became less dominant over time, eventually being replaced by posts from unstable groups around June 2020. This implies that positive posts towards Marcos and Duterte increasingly came from groups with relatively “short-lived” political activity.
We show above the number of stable and unstable groups that were sources of positive posts. Pro-Robredo and pro-Marcos groups were consistently dominated by stable groups. The number of stable pro-Duterte groups remained relatively uniform, but there was an increase in the number of unstable groups. Unstable groups increasingly dominated pro-Marcos groups. Pro-Marcos groups were split almost evenly into stable and unstable groups by June 2020.
Anti-Marcos and anti-Robredo posts shifted from stable groups to unstable groups
Here we show the proportion of negative posts that came from stable groups (red) and unstable groups (yellow) per week. Unstable groups increasingly became the source of negative posts towards Robredo and Marcos, with unstable groups dominating weekly posts by June for Marcos-related posts and August for Robredo-related posts. In other words, more and more anti-Duterte and anti-Robredo groups came from groups with short-lived political activity. Duterte-related posts meanwhile consistently came from stable groups which we may think of as consistent anti-Duterte groups.
We show above the number of stable or unstable anti-Robredo, anti-Marcos, and anti-Duterte groups over time. In general, these groups were dominated by stable groups. However, the number of unstable groups has been increasing over time. In particular, unstable anti-Robredo and anti-Marcos groups have become almost as dominant as stable groups by June 2020.
Some political groups were masked in nonpolitical-sounding names
The top 15 pro-Robredo groups had mostly pro-opposition or pro-Robredo-sounding names. Aside from these, the groups “Netizens for Conscience and Bravery” and “Sobra na! Tama na! Labanan na!” turned out to be pro-Robredo groups.
The groups “Duterte and Marcos Unite Supporters,” “President Duterte News,” and “Rody Duterte for President” contained sarcastic Robredo-related posts which the sentiment model mistakenly tagged as positive, such as “ito ang pinakapani-paniwala na sinabi leni!”, “napakarami mo palang alam madam leni, sana man lang nai-bahagi nyo ito sa ating gobyerno,” and “hahaha pagka brayt gyud ni aleng leni.”
Anti-Robredo groups mostly had names related to Duterte and Marcos. However, other groups with anti-Robredo sentiment were “Tulfo Brothers Worldwide,” “Freedom Wall,” “Political Freedom Supporters,” and “Active Social Media Commentators of the Philippines.” The group “The Resistance” was not an anti-Robredo group, but Robredo was frequently mentioned in posts with negative sentiment pertaining to Marcos or Duterte. Interestingly, “Real Philippine History” turned out to be a dominant anti-Robredo group. This group was also a pro-Marcos group as we will show in the next charts. The groups that drove the bursts of anti-Robredo posts in November 2019 and January 2020 included “Harry Roque sa Senado” and “Real Philippine History.”
Most of the top pro-Marcos groups had Marcos-related names. The group “Real Philippine History” was one of the most popular pro-Marcos groups. This alludes to pro-Marcos “historical revisionism” propaganda that have been known to circulate on Youtube and other social media platforms. The most dominant anti-Marcos groups had diverse names, though many of them share the themes of nationalism and fighting corruption and political dynasties. “Freedom Wall (Original)” was anti-Marcos in contrast to “Freedom Wall,” which was generally anti-Robredo. Note that “Get Real Philippines” was generally pro-Marcos, but the keywords related to Marcos were associated with negative sentiment, especially towards the opposition.
The top pro-Duterte groups mostly had Duterte-related names. “Real Philippine History” and “Tulfo Brothers Worldwide” showed up again to be a popular pro-Duterte group. The top anti-Duterte groups had names that were related to the opposition party, justice, and democracy. The group “Political Freedom Supporters” also had several posts with anti-Duterte sentiment.
Some anti-Robredo posts occurred in bursts or synchronous activity within short periods of time of one month to two. These bursts were not present in trends of anti-Marcos and anti-Duterte posts. These bursts may indicate collective, coordinated action, or may have been triggered by specific political events.
We categorized groups into politically stable groups with consistent political activity, and politically unstable groups whose political activity decayed over time. Pro-Robredo posts were primarily driven by politically stable groups while pro-Duterte, pro-Marcos, anti-Robredo, and anti-Marcos posts became increasingly driven by politically unstable groups. In other words, over time, more and more of these posts came from groups whose political activities were relatively short-lived. Meanwhile, anti-Duterte posts were driven mostly by stable groups.
Finally, we investigated the top groups that drove these political posts and found that some of these groups were masked by non-political-sounding names. Three of the most polarizing groups were “Tulfo Brothers Worldwide,” “Real Philippine History,” and “Harry Roque sa Senado.” “Real Philippine History” and “Harry Roque sa Senado” were two of the groups responsible for the bursts of anti-Robredo posts. – Rappler.com
Prince Javier is a PhD student and teaching fellow at the Asian Institute of Management. Currently, he is a data scientist at Docquity, a startup that develops a knowledge-sharing platform for doctors in Southeast Asia. He is also the co-founder of Bridge 360 IT Solutions, a tech startup in the Philippines that develops technology solutions for small- and medium-scale enterprises as well as for social good.